Review once again manual multiple polynomial regression ! Is it ok to start with a single variable and find its function against the LP. then make multiple regression with another variable then multiple regression with the combining two variables with the remaining one. Answer: No IT IS NOT OK! When you do a multiple regression each time you include the terms for the combining effect of the variables, such as A*sp*m0'. However, if you actually have three variables, you may have such a term B*N*sp*m0' or maybe even more complex version of this instead of something like A*sp*m0' which has only two variables. That's why I need a different strategy:First I will need to linearize each variable by finding the fitting function of a single variable by keeping others constant. Then by modifying the data of each variable by using these functions, I should try first MULTIPLE LINEAR REGRESION.If this does not work as expected than I should try MULTIPLE POLYNOMIAL REGRESSION. Forget about anything else and directly calculate the N - XmM0' multiple reg. but keep in mind that for modified M0 ( M0') I can either Sinus function, or the simple 1st D. Pol. function. In fact, the latter gives a better fit, in the N - XmM0' multiple regression, in contrary to fact that sinus fit gives better regression on M0' regression. Today's tasks: Do not use the. Reinstating the full convergence equation in case. the good memory. The final adjustment needed to be done urgently.